VMware DR solution without Site Recovery

[Ravi1987]

HI Team ,

Note :--- WE are not using SRM.

i have two site which is 1KM away from each other and are conencted by 8 GBPS dark fibre. i have plan to build a New Datacenter with 4 esxi host and 20 VM on Primary and 2 host and 10 VM on DR site. Storage will be DELL Equallogic with 10GBPS conenctivity.

Application(Virtual machine will be running on Primary will be Replicated to DR site by using Equallogic Replication manager software.

Both Primary and DR will be like a Same LAN setup but actually having 1KM physical difference between them.It will be my Near DR.

Design looks ok but i have some queries realted to setup which are mention below .

1. Can i manage both Primary and DR esxi in same cluster since i have a very good Sync Relication and dedicated link between them. Lun which are visble on Primary site will be visible to DR site also.

2. If i am manging both the Primary and DR VMware setup from Same vcenter but having primary and DR two different cluster. Primary lun when relicated to DR storage than how can i mount them in DR esxi server without format them since they will always be having recent data availability in terms of VM.

i will use those vm in case primary fails than i can use them as a primary application.

Thanks in advance.

[sparrowangelste]

1. Can i manage both Primary and DR esxi in same cluster since i have a  very good Sync Relication and dedicated link between them. Lun which are  visble on Primary site will be visible to DR site also.

the latency has to be low, like under 5ms low.

10ms with enterprise plus.

so if you can achive that then then its possible,

I wouls still stick with a different cluster

[Ravi1987]

Yes the letancy will be very low and data change will be approx 50 Gb which will be replicated between primary and DR.

I read we can user Vmware Streched cluster for that kind of setup where letancey is very low and my Distance between sites is only 1 and 1.5 Km only.

i hope after that all the resuorces i can use Vmotion,Svmotion and HA between site successfully.

[jrmunday]

Using a stretched cluster should work in this scenario, but you should take into account the impact cluster features such as HA and DRS. In a stretched cluster with fully automated DRS you will need to create and manage some DRS rules to ensure that you keep your PROD and DR servers in the correct datacenter. Yes, this is a simple task, but Im of the opinion that the solution should be as simple and elegant as possible - ie. don't over engineer something just because you can.

How is your networking setup between the two datacenters? Do you use stretched VLANs (cant remember if you said in your original post) One consideration might be to use separate clusters for simple manageability and have stretched networks configured identically on all of your hosts. This gives you the flexibility to migrate VM's between datacenters and retain the same host name and IP address.

We do this in our environment and have 77km (~40 miles) between sites ... Since the hosts are identical hardware and configurations I can simply vMotion the VM between sites. The hosts are ESXi 5.0 so I have NFS storage presented to both sides which allows me to do this - but I obviously only use this as the stepping stone to get the guest between sites and storage vMotion it onto site specific storage.

[Ravi1987]

thanks forthe reply.

suppose i take dr esxi in different cluster and will manage both primary and dr esxi in two different cluster from same vcenter than can we mount replicated luns in dr esxi host and then run virtual machine from there in case of primary site or cluster fails.

[jrmunday · Accepted solution]

Yes, that's exactly how we do it. vCenter actually runs from the DR site so if production fails, your vCenter infrastructure is already up and running in DR. Since this is a VM as well, you also have the option of migrating it between sites if there is any planned maintenance.

This is also a good example where you should setup some DRS rules to ensure that you always know where your vCenter is in the event of an unplanned outage. I would create a rule which places vCenter on the first host in your cluster (using a should rule, not a must rule) so that you can simply point your vSphere client at this host and bring up vCenter first when you need to - this saves checking each host to find out where it is.

[Ravi1987]

Thanks for providing Such a detailed answer.

Please let me know one last thing that by using SRM we can automate the procedure of  mounting Replicated luns from primary to DR site without causing any Data on any Lun.

Can we do the same without using SRM. my motive is to replicate all the production lun to DR site and mount them on DR esxi server's.

In case of primary down i cn simple power on the virtual machines... can i do this.